A New Encoding for the Degree Constrained Minimum Spanning Tree Problem

Abstract

We present an effective new encoding method for use by black-box optimisation methods when addressing tree-based combinatorial problems. It is simple, easily handles degree constraints, and is easily extendable to incorporate problem-specific knowledge. We test it on published benchmark degree-constrained minimum spanning tree (DC-MST) problems, comparing against two other well-known encodings. The new method outperforms the comparative encodings. We have not yet compared against the recently published 'edge-sets' encoding, however we can report preliminary work which indicates sophisticated versions of the new encoding can outperform edge-set on at least some classes of DC-MST. © Springer-Verlag 2004.